Please ensure Javascript is enabled for purposes of website accessibility

Md. firm uses AI, algorithms to match manufacturers with hard-to-find parts

Randy Altschuler says Xometry uses artificial intelligence and algorithms to help manufacturers track down niche parts. MUST CREDIT: Washington Post photo by Michael Robinson Chavez.

Randy Altschuler says Xometry uses artificial intelligence and algorithms to help manufacturers track down niche parts. (Michael Robinson Chavez/The Washington Post)

When NASA needed $59,000 in custom components so its International Space Station would have clean air, an engineer went online, punched in the specifications and hit send.

On the receiving end, a Maryland company put its patented software to work and within seconds had a price and delivery date for parts destined to live 250 miles beyond Earth.

Gaithersburg-based Xometry helps companies fill supply chain gaps by quickly combing through thousands of providers to track down custom parts.

“We combine artificial intelligence with algorithms to predict the market price of customized parts within seconds,” Xometry co-founder and chief executive Randy Altschuler said. “The same software then scours our 5,000 manufacturers to deliver the right source at the right price.”

Altschuler said revenues have been growing fast: Sales topped $800,000 in 2014, $2.9 million in 2015, $80 million in 2019 and are now headed toward $150 million. He wouldn’t get into details about the profit side, other than to say he expects the company will turn one “in the next couple of years.”

But its gross margin before payroll and expenses is probably at least $30 million.

Xometry has 400 people on its payroll, including software engineers and data scientists. Half the staff works out of Xometry’s three Washington-area offices. The rest are scattered across the United States and Europe.

Altschuler, who would not say how much of the company he still owns, said he and co-founder Laurence Zuriff plan to take Xometry public in 2021. They have hired a chief financial officer and added former Washington Post publisher Katharine Weymouth to its board of directors in preparation.

Since launching in 2014, Xometry has raised around $150 million from such investors as T. Rowe Price Funds, Henry Ellenbogen’s Durable Capital Partners, the Foundry Group, Highland Capital Partners and Dell Technologies Capital.

“There are thousands of small manufacturers across the U.S. making small things and doing billions of dollars of business,” Altschuler said. “It’s very fragmented, mostly family businesses doing a couple of million dollars a year. It could be a couple of brothers running a manufacturing shop started by their grandfather.”

These small suppliers have been heavily reliant on the health of big companies in their regions. If you are in Houston, your business lives and dies with oil and gas. In the Midwest, it’s automotive. Boston caters to medical device companies. Washington is well known for its aerospace and defense.

Altschuler has essentially created on-demand shopping for the $260 billion business-to-business marketplace for custom manufactured parts.

“There’s no directory to find these small manufacturers,” he said. “There is no Craigslist. No place for these guys to get work when those local big industries aren’t doing well.”

Xometry has opened up these small companies to big firms like Bosch, General Electric, Dell and BMW, all of which may be looking for a just-in-time piece like an injection-molded part or metal 3-D printing that might be a few hundred or few thousand dollars – not enough to buy in huge quantities. Xometry’s average sale is $2,000 to $3,000.

Xometry’s artificial intelligence sifts through thousands of manufacturers in its database and zeros in on the one best suited to make the part and at a price that allows Xometry to turn a profit.

Its gross margin – known as the “take rate” – comes from the difference between what the online shopper such as NASA pays for a part and what the small manufacturer sells it for.

“We are on the hook for finding a price that is cheaper than what we have already sold the part for,” Altschuler said.

The program determines which company gets first crack at the job. If the supplier doesn’t respond, a message is sent to the next-best candidate, and the process keeps going until there is a confirmed match. Xometry makes money on more than 90 percent of the transactions.

“A manufacturer might be sipping a Coke at McDonald’s when he or she presses their iPhone and agrees to the job,” Altschuler said, adding that 30% of orders are accepted on cellphones.

Altschuler said there was no “aha moment” that spawned the idea for Xometry – “We just figured it out.”

He is a self-described “grinder” who was raised by his mother in New York. He worked his way through Princeton as a cook and security guard and spent a year in Vienna as a Fulbright scholar before heading to Wall Street. He worked at private equity firm Blackstone before striking out on his own with a college buddy.

He started his first company, OfficeTiger (after the Princeton mascot), in 2000. He sold it six years later for $250 million. He was 35 years old.

“I was fortunate,” Altschuler said.

Still ambitious, he squeezed in a run for Congress as a Republican on Long Island. He came close, but lost in a recount by a few hundred votes.

In 2008 he co-founded CloudBlue, an electronics recycling firm that helped large companies dispose of data-sensitive equipment. He sold it five years later for around $40 million.

Altschuler said his entrepreneurism comes from an innate curiosity and a love for research and networking. He works through ideas with friends, former co-workers and others he finds through the alumni list of Harvard Business School, from which he graduated in 1998. He also works connections through Washington-area business associations.

“I get interested in something and spend a lot of time studying,” said Altschuler, who lives with his wife and children in Potomac. “I meet with companies. I reach out to folks. I know what I know and know what I don’t know.

To purchase a reprint of this article, contact